The Wage and Employment Impact of Minimum-Wage Laws in Three Cities

8/7/2019 The Wage and Employment Impact of Minimum-Wage Laws in Three Cities

1/34

TheWageandEmployment

ImpactofMinimumWageLaws

inThreeCitiesJohnSchmittandDavidRosnick

March

2011

CenterforEconomicandPolicyResearch

1611ConnecticutAvenue,NW,Suite400

Washington,D.C.20009

2022935380

www.cepr.net


2/34

CEPR The Wage and Employment Impact of Minimum-Wage Laws in Three Cities i

Contents

Executive Summary........................................................................................................................................... 1

Introduction........................................................................................................................................................ 2Estimation and Data.......................................................................................................................................... 3

Estimation ...................................................................................................................................................... 3

Data................................................................................................................................................................. 5

Minimum-Wage Bite ..................................................................................................................................... 6

Results.................................................................................................................................................................. 9

San Francisco................................................................................................................................................. 9

Santa Fe......................................................................................................................................................... 22

Washington, DC.......................................................................................................................................... 24Discussion......................................................................................................................................................... 24

Conclusion ........................................................................................................................................................ 29

References......................................................................................................................................................... 31

About the Authors

John Schmitt is a Senior Economist and David Rosnick is an economist at the Center for Economicand Policy Research, in Washington, D.C.

AcknowledgmentsThe Center for Economic and Policy Research is grateful to the Russell Sage Foundation and theInstitute for Research on Labor and Employment at the University of California at Berkeley for theirfinancial support. The authors thank the Bureau of Labor Statistics for access to and assistance withthe confidential Quarterly Census of Employment and Wage data. We also thank Arindrajit Dube,Michael Reich, Dean Baker, Eileen Appelbaum, Michele Mattingly, and Heather Boushey for manyhelpful comments and suggestions.


3/34

CEPR The Wage and Employment Impact of Minimum-Wage Laws in Three Cities 1

ExecutiveSummary

This report analyzes the wage and employment effects of the first three city-specific minimum wagesin the United States San Francisco (2004), Santa Fe (2004), and Washington, DC (1993). We usedata from a virtual census of employment in each of the three cities, surrounding suburbs, andnearby metropolitan areas, to estimate the impact of minimum-wage laws on wages and employmentin fast food restaurants, food services, retail trade, and other low-wage and small establishments.

We evaluate these impacts using the general approach outlined in Card and Kruegers (1994, 2000)studies of the 1992 New Jersey state minimum-wage increase. In our setting, the Card and Kruegermethodology involves comparing wages and employment before and after the city minimum-wage with changes over the same period in wages and employment in comparable establishments innearby areas unaffected by the citywide minimum wage.

The results for fast food, food services, retail, and low-wage establishments in San Francisco andSanta Fe support the view that a citywide minimum wages can raise the earnings of low-wage

workers, without a discernible impact on their employment. Moreover, the lack of an employmentresponse held for three full years after the implementation of the measures, allaying concerns thatthe shorter time periods examined in some of the earlier research on the minimum wage was notlong enough to capture the true disemployment effects.

Our estimated employment responses generally cluster near zero, and are more likely to be positivethan negative. Few of our point estimates are precise enough to rule out either positive or negativeemployment effects, but statistically significant positive employment responses outnumberstatistically significant negative elasticities.

In general, our findings are consistent with earlier research by Card and Krueger (1994, 1995, 2000),

Dube, Lester, and Reich (2011), and others who have found at the federal and state level thatpolicy makers tend to set levels of the minimum wage that raise wages of low-wage workers withoutsignificant negative effects on the employment of low-wage workers. Our findings are alsoconsistent with earlier studies of the citywide minimum wages in San Francisco (Dube, Naidu, andReich, 2006) and Santa Fe (Potter, 2006).

We also find that the minimum-wage increase implemented in Washington, DC, in 1993 was toosmall to raise wages in fast food, food services, retail, and other low-wage establishments. Data forWashington show that a relatively low share of the citys hourly workers were in the wage rangeaffected by the new city minimum and also suggest that the measure was not enforced (moreworkers earned below the new city minimum after it was enacted than had earned in that rangebefore the law went into effect). The citywide minimum wage in Washington therefore does notconstitute a meaningful policy experiment, and does not allow us to draw conclusions about theemployment effects of binding citywide minimum wages.

The experience of smaller establishments in San Francisco and Santa Fe whether they were initiallybelow or just above the size exempted by the new laws suggests that small establishments do notrespond to minimum wages differently than larger firms. Small establishments below the size cutoffsin those cities did not experience systematic changes in employment. The same was true for smallestablishments just above the size cutoffs in these cities.


4/34


Introduction

The United States has extensive experience with minimum-wage legislation. In 1938, the Fair LaborStandards Act established a national minimum wage, which has been raised periodically ever since(most recently in 2007, 2008, and 2009). Beginning in the 1980s, and especially in the 1990s and2000s when the federal minimum wage spent long periods at the same nominal level, more than halfof U.S. states enacted state-level minimum wages above the federal level. Most recently, four cities Washington, DC (1993), San Francisco (2004), Santa Fe (2004), and Albuquerque (2007) implemented city-wide minimum wages at levels set above the prevailing national or state minimumwages.

Economists have produced a large body of research on the economic impacts of the federal andstate minimum-wage laws,1 but little research analyzes the impact of citywide minimum wages. 2 Theeconomic effects, especially on employment, of city-specific minimum wages could differ fromnational and state minimum wages because the much smaller geographic area covered by city lawsmight make it easier for employers affected by the minimum wage to relocate employment outside

of city boundaries; citywide minimum wages might also act to reduce the product-marketcompetitiveness of covered firms relative to uncovered firms in nearby suburbs or cities.

In this paper, we analyze the wage and employment effects of the first three city-specific minimum wages Washington, DC, San Francisco, and Santa Fe.3 We use data from a virtual census ofemployment in each of the three cities, surrounding suburbs, and nearby metropolitan areas, toestimate the impact of each citys minimum-wage law on wages and employment in small businesses,low-wage establishments, and several low-wage industries, including fast food restaurants and retailtrade. We evaluate these impacts using the general approach outlined in Card and Kruegers (1994,2000) studies of the 1992 New Jersey state minimum-wage increase. In our setting, the Card andKrueger methodology involves comparing wages and employment before and after the city

minimum-wage with changes over the same time frame in wages and employment in comparableestablishments in nearby areas unaffected by the citywide minimum wage.

Our research extends earlier minimum-wage research in several dimensions. First, we focus narrowlyon the effects of citywide minimum wages, a new area of policy innovation where research is thin.Second, we present side-by-side analysis of the first three citywide minimum wages, including thefirst look at the Washington, DC citywide minimum wage. Third, we complement Dube, Naidu, andReichs (2006) analysis, based on their own survey of restaurants in San Francisco and nearby areas,

1 For reviews of the empirical research on the minimum wage, see Brown, Gilroy, and Kohen (1982), Card and Krueger(1995), Brown (1999), Neumark and Wascher (2006), and Flinn (2011).

2 Dube, Naidu, and Reich (2006) conducted an empirical analysis of the 2004 San Francisco minimum wage based on

their own survey and other data. Potter (2006) analyzed microdata from the New Mexico state unemploymentinsurance system to study the 2004 Santa Fe increase. Dube, Kaplan, Reich, and Su (2006) have written a summary ofthe available research on San Francisco and Santa Fe. To our knowledge, no comparable research covers the 1993

Washington or the 2007 Albuquerque minimum wage. Card and Krueger (1994, 2000) and Dube, Lester, and Reich(forthcoming) focus on local discontinuities in the minimum wage across state borders, but dont specifically addresscitywide minimum wages.

3 On January 1, 2007, Albuquerque implemented a citywide minimum wage set at $6.75 per hour, increasing to $7.15 perhour on January 1, 2008, and then to $7.50 per hour on January 1, 2009 (with a lower rate for employers providinghealth-insurance benefits above a $2,500 per year threshold). Data lags and the desire to analyze effects over threeyears after the increase have forced us to exclude Albuquerque from this analysis.


5/34


with data from the Bureau of Labor Statistics (BLS) Quarterly Census of Employment and Wages(QCEW) program, which is a virtual census of employment conducted in connection with state-levelunemployment insurance systems. Fourth, we look beyond the traditional work horses of theminimum-wage literature fast-food restaurants and food services more broadly to other low- wage industries, including retail trade, and to low-wage establishments, independent of industry.

Fifth, we also examine the impact of citywide minimum wages on establishments of different sizes,allowing us to evaluate the particular impact on small firms. Sixth, in each case, we are able to trackemployment and wage changes for three full years after implementation, permitting us to addressconcerns that the one-year (and shorter) post-increase evaluation period in many earlier studies maynot be long enough to capture the full economic impacts. Finally, in recognition of the potentialsensitivity of difference-in-difference estimators to the control group chosen, for each city we reportresults from three different control groups: the citys suburbs, a nearby city, and a nearby city andthat citys suburbs.

Overall, we find little evidence that the three citywide minimum wages had any systematic effect onemployment in low-wage establishments, including the fast-food industry, the broader food-servicessector, and retail trade. Our estimated minimum-wage policy elasticities generally cluster near zero,

and are slightly more likely to be positive than negative. Relatively few of our point estimates areprecise enough to rule out either positive or negative employment effects, but more of the positiveelasticities are based on underlying statistically significant employment and wage effects than are thenegative elasticities. These findings on the lack of discernible employment effects hold for threeyears after the implementation of the minimum wage and across the three different geographiccontrol groups we use to benchmark the minimum-wage effects.

EstimationandData

Our approach draws heavily on Card and Kruegers (1994, 2000) methodology. In general, wecompare establishments affected by the citywide minimum wage with establishments in nearby areasnot covered by the citys new minimum wage. 4 As in Card and Krueger (2000), we use confidentialdata drawn from each states unemployment insurance system, which covers about 98 percent of allemployees.

Estimation

We take the city minimum wage (seeTable 1) as exogenous and designate establishments coveredby each of the new laws as three separate treatment groups. For each of our three cities, we alsodefine three separate control groups, chosen for their geographic proximity: the treatment cityssuburbs, a nearby city, and a nearby city and its suburbs.5

4 Since both San Francisco (during a phase-in) and Santa Fe (indefinitely) exempted smaller firms from their citywideminimum wages, we can also compare wage and employment outcomes across firms of different sizes within the samegeographic area. In San Francisco, establishments with fewer than 10 employees or that operated as nonprofitorganizations were not covered by the citywide minimum-wage law for the first two years it was in effect. In Santa Fe,firms with fewer than 25 employees are not covered.

5 For San Francisco, we define the suburbs as Marin, San Mateo, and San Francisco counties; the control city asOakland; and the Oakland suburbs as Alameda and Contra Costa counties. For Santa Fe, the suburbs are Los Alamosand Santa Fe counties; the control city, Albuquerque; and the Albuquerque suburbs, Bernalillo, Sandoval, and Valenciacounties. For Washington, the suburbs are Montgomery and Prince Georges counties in Maryland and Arlington and


6/34


TABLE 1

City Minimum-wage Laws

San Francisco Santa Fe Washington, DC

Entered into effect February 23, 2004 June 22, 2004 October 1, 1993

Initial rate $8.50 $8.50 $5.25

Subsequent increases Indexed to inflation Scheduled increases to$9.50 on January 1, 2006,and $10.50 on January 1,2008; the $10.50 increase

was not implemented,instead at the end of 2007,

the city council voted toindex future rate increases

to inflation

Federal minimumwage plus $1

Exemptions Two-year phase infor establishments

with fewer than 10employees and fornon-profit 501(c)(3)

organizations

Establishments with fewerthan 25 employees exempt;

at the end of 2007, the citycouncil voted to extendcoverage to all

establishments, regardlessof size

None

Following Card and Krueger (1994, 2000), Dube, Naidu, and Reich (2006), Potter (2006), andothers, we use a simple (regression-adjusted) difference-in-differences estimator.6 For different typesof establishments (by industry, by initial employee-size, or average initial wage level), we take datafor each establishment, j, in treatment city, i, and calculate (separately) the change in our twovariables of interest, y (average establishment wage or total establishment employment), before and

after the implementation of the minimum-wage. We then compare these changes in the treatmentgroup with analogous changes in the same variables of interest in each of the three control groups,k:

E(yij,t+1 - yij,t | Treatmentij=1) - E(yij,t+1 - yij,t | Treatmentij=0, Controlijk=1)

where k = 1 own suburbk = 2 control cityk = 3 control city and control suburb

In the case of the third control group, the estimator is actually a difference-in-difference-in-differences estimator, since we first compare the treatment city before and after the minimum wage

with the surrounding (untreated) suburbs before and after the minimum wage, and then comparethis result with a comparison of the (untreated) control city before and after the minimum wagerelative to the (untreated) suburbs of the control city over the same period. Effectively, we are

Fairfax counties and the cities of Alexandria, Fairfax, and Falls Church in Virginia; the control city, Baltimore; and theBaltimore suburbs, Anne Arundel, and Baltimore counties.

6 See Bertrand, Duflo, and Mullainathan (2004) and Dube, Lester, and Reich (forthcoming) for critiques of difference-in-difference estimators.


7/34


asking how the treatment city performed relative to its suburbs, using how the control cityperformed relative to its suburbs as a benchmark.

In the tables reported below, we implement this framework in a regression context that incorporates(where appropriate) a variety of control variables including establishment size and industry.7

As in earlier research, we report results for fast-food restaurants (see Card and Krueger, 1994, 2000;Neumark and Wascher, 2000; Dube, Naidu, and Reich, 2006; and others) and food services (Dube,Naidu, and Reich, 2006). The QCEW data, however, also allow us to analyze the wage andemployment impacts of citywide minimum wages on other low-wage industries including retailtrade.8

With the QCEW data, we can also conduct similar analyses of changes in employment and wages byinitial establishment size. This feature of the data lets us examine concerns that citywide minimumwages might have a particularly negative impact on small firms. Similarly, we can use the QCEWdata to focus on the impact of the minimum wage on all low-wage establishments within each city,regardless of the establishments industry.

Data

We analyze confidential data from the Quarterly Census of Employment and Wages, a datasetconstructed and maintained by the BLS from administrative data gathered by each of the state-levelunemployment insurance systems. Since the unemployment insurance system covers about 98percent of all employees (agricultural workers are the principal group missing from the dataset), theQCEW data constitute a virtual census of employment, providing large samples even at the locallevel.

For each calendar quarter, the QCEW shows each establishments average total employment (at one

point in time during each of the three months). The dataset also gives the total wages and salariespaid out to these employees over the full quarter. In addition, the QCEW reports detailedinformation on the establishments location and main industry.

The main features of the QCEW are its large size and nearly universal coverage, which allows for afine-grained analysis of the economic impacts of the minimum wage, even at the local level. Thedata set, however, has several important limitations.

First, the QCEW contains no information on hourly earnings. The state unemployment insuranceagencies gather and report data on total employment and on the total wages and salaries paid overthe quarter. These data allow us to calculate an average quarterly employment level for theestablishment and an average quarterly (or monthly) wage per worker, but do not allow us to

calculate an average hourly wage rate.

7 The two dependent variables are the differences in the natural logarithm of the establishments average quarterly wageand quarterly average total employment (plus one, to deal with establishment births and deaths). The regressions areestimated using ordinary least squares (with robust standard errors) and weighted by average establishment size overthe full period covered in each regression.

8 Potter (2006) also uses state unemployment insurance data from New Mexico to analyze the employment impacts ofthe Santa Fe minimum wage on a variety of industries.


8/34


Second, reporting errors and inconsistencies create problems with both identifying the geographicallocation of some establishments and with tracking the establishments over time. A complete set ofprograms used to identify and match establishments are available upon request.

Third, since the QCEW is compiled from self-reported administrative data, the dataset does not

automatically track births and deaths of establishments. Since the birth of a new establishment or thedeath of an old establishment after the implementation of the minimum wage could have animportant impact on overall employment in the treatment and control areas, we have createdrecords with zero employment and missing values for wages for establishments that were born ordied during the course of our analysis. For example, if an establishment existed in the year beforethe minimum wage, but closed down in the year after the minimum wage, the QCEW would containno record for the establishment after the minimum wage. For this analysis, we create a record forthe establishment with total employment of zero and a missing value for wages and link therecord to the same establishment before the minimum wage.

Finally, the QCEW data are confidential and we are restricted in the results we can report. In orderto comply with confidentiality requirements, we have limited what we present here largely to

regression coefficients (and to regressions with thirty or more degrees of freedom).

MinimumWageBite

All three minimum wages were high relative to the previously prevailing state or federal minimumwage.Table 2 shows the size of the minimum-wage policy treatment for each of the three cities. The first column shows the minimum wage prevailing in each city before and after theimplementation of the minimum wage. At each point in time, the prevailing minimum wage in eachcity is the maximum of the applicable federal, state, or citywide minimum wage. The second column

shows the prevailing minimum wage at the state level, which is the maximum of the federal or anystate minimum wage. The third column reports the federal minimum wage. (Wherever minimum wage laws changed in the middle of a period, the reported minimum wage is the time-weightedaverage of the nominal value of the minimum wage over the period.)


9/34


TABLE 2

Nominal Value of Minimum Wage Applicable at the City, State, and National Levels

Differential

Period City State Federal (percent)

(a) San Francisco

2003:Q1 - 2003:Q4 6.75 6.75 5.15 0.0

2004:Q1 8.50 6.75 5.15 25.9

2004:Q2 - 2005:Q1 8.53 6.75 5.15 26.4

2005:Q2 - 2006:Q1 8.67 6.75 5.15 28.4

2006:Q2 - 2007:Q1 8.90 6.94 5.15 28.3

(b) Santa Fe

2003:Q2 - 2004:Q1 5.15 5.15 5.15 0.0

2004:Q2 8.50 5.15 5.15 65.0

2004:Q3- 2005:Q2 8.50 5.15 5.15 65.0

2005:Q3 - 2006:Q2 9.00 5.15 5.15 74.8

2006:Q3 - 2007:Q2 9.50 5.15 5.15 84.5

(c) Washington, DC

1992:Q4 - 1993:Q3 4.25 4.25 4.25 0.0

1993:Q4 5.25 4.25 4.25 23.5

1994:Q1 - 1994:Q4 5.25 4.25 4.25 23.5

1995:Q1 - 1995:Q4 5.25 4.25 4.25 23.5

1996:Q1 - 1996:Q4 5.38 4.38 4.38 22.9

Notes: See Table 1 for coverage details. Date in bold indicates effective date for implementation.Differential between city and surrounding minimum wage is with respect to larger of federal orstate minimum wage. Albuquerque, which is the control city for Santa Fe had a minimum wageof $6.75 per hour from January 1, 2007 through the end of the period covered in this analysis.

The final column in the table calculates the difference (in percent terms) between the minimumwage in each city and the binding (highest) minimum wage in the surrounding state (or, in the caseof Washington, DC, the federal minimum wage). The San Francisco law increased the minimumwage there in January 2004 to $8.50 per hour, about 26 percent above the California state minimumwage of $6.75. The June 2004 minimum wage9 in Santa Fe raised the applicable minimum wage to$8.50, an increase of 65 percent from the federal minimum of $5.15. With subsequent increases inSanta Fe, the gap between the city and the federal minimum wages rose over three years to over 80percent. The implementation in October 1993 of the citywide minimum wage in Washington, DC,to a level $1.00 above the prevailing federal minimum wage of $4.25 increased the legal minimumwage for Washington establishments by about 24 percent. 10

9 Since the June 22, 2004 increase did not go into effect until the last week of the second quarter after the monthlyreporting period for the QCEW employment numbers our analysis treats the third quarter of 2004 as the effectiveimplementation date. We also produced identical estimates using the effective implementation date in the firstquarter (when the law was originally supposed to go into effect until blocked by a court order) and the secondquarter (which includes the actual implementation date of June 22), but found results qualitatively similar to thosepresented below (results available upon request).

10 Neither of the two states bordering Washington Maryland and Virginia had a state minimum wage.


10/34


An increase in the prevailing legal level of the minimum wage is a necessary, but not sufficient,condition for the city minimum wage to bite. An increase in the legalminimum wage would haveno impact on wages if the new minimum were set below what workers in the citys low-wage labormarket were already receiving. In addition, even if a city has a substantial pool of workers earningbelow the new minimum wage, the law might have no effect on wages if it is not enforced.

Table 3 reports results of a simple analysis of the initial, overall bite of the city minimum wages ineach of the three cities. The table uses data on hourly paid workers from the Current PopulationSurvey (CPS) in the twelve months immediately before each city minimum wage went into effect.The CPS data have severe limitations in this context. Technically speaking, the sub-state data for SanFrancisco and Santa Fe are not representative, since the CPS is not designed to provide validsamples at the sub-state level. The data for Santa Fe have the additional limitation that the surveydoes not allow us to distinguish between central city and suburban residents; and for SanFrancisco the data after April 2004 do not distinguish between San Francisco and other nearbyareas, including Oakland (and, therefore, observations from May 2004 are not included in panel (b)of Table 3). But, the most serious limitation and the main reason we turn below to the QCEWdata is that the sample sizes for San Francisco and Santa Fe are very small. For Washington

however, the sample is technically valid, because the city is treated as a state in the CPS. The Washington sample is also small, but can give a rough a picture of the citys wage distribution,especially relative to the nation as a whole.

TABLE 3

Share of Hourly Paid Employees Affected by City Minimum Wage

United States

excluding

San Francisco Santa Fe Washington, DC Washington, DC

(a) Before Feb 2003-Jan 2004 Jun 2003-May 2004 Oct 1992-Sep 1993 Oct 1992-Sep 1993

Below range (%) 3 3 2 3In range (%) 13 18 10 19

Above range (%) 84 79 88 79

Sample size 131 76 684 105,473

(b) After Mar 2004-Apr 2004 Jul 2004-Jun 2005 Nov 1993-Oct 1994Nov 1993-Oct 1994

Below range (%) 0 0 2 3

In range (%) 6 19 14 17

Above range (%) 94 81 84 80

Sample size 17 85 639 104,080

Notes: Authors' analysis of National Bureau of Economic Research extracts of the Current PopulationSurvey Outgoing Rotation Group files. Weighted share of hourly paid workers earning (1) less than thebinding federal or state minimum wage prevailing in the 12 months before the city minimum wage("Below range"); (2) at least the binding federal or state minimum wage but less than the cityminimum wage ("In range"); and (3) at least the city minimum wage (Above range). From May2004, the city of San Francisco cannot be distinguished in the public CPS data from Oakland or SanJose and therefore these observations are excluded from the calculations. The CPS data do notdistinguish between "central city", suburbs, and other areas in the Santa Fe metropolitan area.


11/34


The CPS data, while too limited to be definitive, suggest that the minimum wage in Washington wasnot binding. The city, relative to the nation as a whole, had high wages before the city minimumwage went into effect, and the new minimum was set at a level that affected relatively few workers.In the year before the increase went into effect, only about 10 percent of the citys paid hourlyworkers earned somewhere between the $4.25 per hour set by the federal government and the $5.25

imposed by the new law (see panel (a) of Table 3). By comparison, about 19 percent of workers inthe United States were in this pay range. More importantly, the CPS data suggest that the higher cityminimum was not enforced. While the change in coverage was not statistically significant, the shareof workers earning less than the new minimum wage actually increased (to about 14 percent) in thetwelve months after the minimum wage went into effect. If the minimum wage had been binding,and enforced, we would have expected the share of workers earning between $4.25 and $5.25 perhour to have declined not increased after the new law went into effect.

These CPS data provide some context for the results we find below with the much larger QCEWsample. While we find many cases where the city minimum wage increased the average wage in fastfood, food services, retail, and low-wage establishments in San Francisco and Santa Fe, we do notfind a single case where the Washington minimum wage was associated with higher wages in these

same typically low-wage establishments. We conclude from the CPS and QCEW data that theminimum-wage increase enacted in Washington was not large enough to constitute a reasonable testof the employment impacts of the minimum wage. For completeness sake, however, in the rest ofthe paper, we report our results for Washington alongside those of San Francisco and Santa Fe.

Results

In this section, we review, city-by-city, the results of our analysis. For each city, we begin with thefast-food industry and food services more broadly, then look at retail trade, another typically low-

wage industry. We then step back and look at outcomes in low-wage establishments (those in thecitys bottom fifth of average pay), regardless of industry. Finally, given concerns that minimum-wage laws may have a negative impact on smaller firms, we study the impact of the minimum wagelaws on firms by establishment size.

SanFrancisco

Table 4 reports our estimates of the impact of these minimum-wage laws on wages andemployment in fast-food restaurants. Panel (a) presents our estimates of the impact on the averagewage in fast-food restaurants. For each city, we give nine estimates: the effect of the new law afterone, two, and three years, using each of the three controls: own suburbs, control city, control cityand suburbs. According to our estimates, the 26 to 28 percent increase in the prevailing minimum

wage in San Francisco appears to have raised the average fast-food wage there 4 to 5 percent11 in thefirst year after the law was enacted, relative to the three geographic benchmarks where the minimumwage did not increase. The estimated wage impact was generally larger over time 5 to 11 percent in

11 The coefficients in Tables 3 through 9 are log points. A 0.1 log point change is approximately equal to a 10 percentchange. For larger changes, the log-point measure understates the percent change. For ease of exposition, we convertthe log-point changes to percent changes assuming rough equivalence.


12/34


the second year and 9 to 11 percent in the third year. 12 Six of the nine estimates are statisticallysignificant at the five percent level or better; the other three are statistically significant at the tenpercent level. Taken together, these results suggest that the San Francisco law had an economicallyand statistically significant impact on the wages paid in fast-food restaurants.

TABLE 4

Wage and Employment Effects of City Minimum-wage Laws, Fast Food Restaurants


Other Other Other Other Other Other

Own nearby cities & Own nearby cities & Own nearby cities &

Control group suburbs city suburbs suburbs city suburbs suburbs city suburbs

(a) Change in ln (average establishment wage)

One year 0.05 0.04 0.04 0.08 0.03 0.03 0.01 0.01 -0.01

(0.01)** (0.02)* (0.02)# (0.04)* (0.02) (0.05) (0.02) (0.02) (0.03)

Two years 0.11 0.05 0.08 0.09 0.09 0.03 -0.04 -0.03 -0.02

(0.03)** (0.03)# (0.04)# (0.04)# (0.03)** (0.05) (0.03) (0.04) (0.04)

Three years 0.11 0.09 0.09 0.01 0.07 -0.06 -0.06 -0.05 -0.06(0.02)** (0.03)** (0.03)** (0.11) (0.03)* (0.12) (0.03)* (0.03) (0.04)#

(b) Change in ln (average establishment employment)

One year -0.11 -0.01 -0.04 0.24 -0.03 0.38 -0.20 -0.16 -0.10

(0.07)# (0.10) (0.12) (0.16) (0.09) (0.18)* (0.06)** (0.06)* (0.08)

Two years 0.08 -0.07 0.05 0.31 -0.12 0.49 -0.43 -0.16 -0.17

(0.18) (0.13) (0.24) (0.21) (0.11) (0.23)* (0.14)** (0.12) (0.16)

Three years 0.05 -0.072 0.03 0.36 -0.08 0.61 -0.56 -0.35 -0.30

(0.20) (0.15) (0.25) (0.21)# (0.13) (0.24)* (0.16)** (0.15)* (0.19)

(c) Labor-demand elasticities

One year -2.3 -0.2 -1.0 2.8 -0.9 12.8 -26.2 -26.5 9.7Two years 0.7 -1.3 0.7 3.7 -1.3 16.6 11.9 5.8 8.1

Three years 0.5 -0.8 0.4 63.9 -1.2 -10.5 9.2 7.4 5.1

(d) Employment-to-minimum-wage elasticities

One year -0.4 0.0 -0.1 0.4 0.0 0.7 -0.7 -0.5 -0.4

Two years 0.3 -0.2 0.2 0.5 -0.1 0.8 -1.5 -0.6 -0.6

Three years 0.2 -0.2 0.1 0.5 -0.1 1.0 -1.9 -1.3 -1.1

Notes: Authors' estimates using QCEW data. Ordinary least squares; robust standard errors in parentheses: **, 1percent; *, 5 percent; #, 10 percent. All regressions include controls for firm size (10-24, 25-99, 100-999 and1,000 or more).

The wage estimates in panel (a) suggest that we have a meaningful policy experiment: the minimumwage appears to have increased wages substantially in San Francisco. Panel (b) of Table 4 presentscorresponding estimates of the employment impact of the minimum-wage laws. For San Francisco,all three of the one-year estimates are negative. The smallest two of these estimates (-1 to -4 percent)are not statistically significant; the largest (-11 percent) is statistically significant at the 10 percent

12 Adjustment costs could plausibly lead employers to phase in employment changes in the wake of a minimum wage; itis more difficult to explain how adjustment costs might affect wage changes over time.


13/34


level. None of the two- and three-year estimates, however, are different from zero at standard levelsof statistical significance (four of these results are positive, two are negative).

Taken together, the results for San Francisco in panels (a) and (b) give little support to the view thatthe minimum wage there had a negative effect on employment in fast food. While wages rose by an

economically and statistically significant amount in nine of the nine policy experiments, employmentdid not change by a statistically significant amount in eight of those nine cases (and was onlystatistically significant at the ten percent level in the remaining case).

Panels (c) and (d) present these wage and employment findings more formally. Panel (c) calculatesthe standard labor-demand elasticity:

% Employment% Wage

calculated as the ratio of entries in Panel (b) roughly the percent change in employment to thecorresponding entry in Panel (a) roughly the percent change in wages. Panel (d) presents the

elasticity that is commonly reported in the minimum-wage literature, which relates the percentchange in employment in a subpopulation (here, fast-food workers) to the percent change in theminimum wage:

% Employment in Subpopulation% Minimum Wage

Focusing on the policy elasticities in panel (d), the minimum wage was associated with employmentincreases in four cases, zero change in one, and employment declines in four cases. Eight of the nineestimates were narrowly distributed in the range between -0.3 and 0.3, consistent with most earlierresearch (see Figure 1). Only one of the nine cases was based on a statistically significant (at the 10percent level) employment decline: in the first year after the increase, using the San Francisco

suburbs as a benchmark. The estimated policy elasticity in that case was -0.4.


14/34


FIGURE 1

Employment-to-minimum-wage Elasticities, Fast Food Restaurants(a) San Francisco

c s

S c b c b b s

-1.2 -.9 -.6 -.3 0.0 .3 .6 .9 1.2

(b) Santa Fe

c

c s s

-1.2 -.9 -.6 -.3 0.0 .3 .6 .9 1.2

Notes: These histograms summarize all of the cases where the minimum-wage legislation was associated with anincrease in the average establishment wage in the industry (where the increase was statistically significant at at leastthe 10 percent level). If the minimum wage consistently lowered employment, then the estimated policy elasticitieswould all be to the left of the bar above zero. Elasticities based on own suburbs labeled S; reference city, C;reference city and its suburbs, B. Elasticities for year one, in red; year two, in blue; year three, in black. Elasticities inlower-case are not statistically significantly different from zero; estimates in upper case are significant at the 10percent level; in upper case and bold, at the 5 percent level or better. Estimates based on statistically insignificantincreases in average wages, or declines in wages, not shown. A maximum of nine possible elasticity estimates percity.Source: See Table 4.

Table 5 presents the results from an identical analysis for the broader category of food services, which includes fast food, but also sit-down restaurants. As with fast-food restaurants, the SanFrancisco minimum wage led to a statistically significant increase in food service wages relative to allthree control groups in all three years (see panel (a)). By our estimates, the average wage in SanFrancisco food services increased 3 to 7 percent in the first year, about 6 percent in the second year,and about 7 percent in the third year, relative to the three geographic benchmarks where theminimum wage did not increase. Over the same period, food service employment in San Franciscoincreased relative to employment in all three geographic control groups in all three years (see panel(b)). In two of these cases, the employment increases were statistically significant at the five percentlevel; in the third case, at the ten percent level. The estimated employment changes appeared to bestrongly tied to the geographic control. Employment changes were smallest relative to the immediate

San Francisco suburbs (1 to 6 percent); larger relative to the nearby control city of Oakland (15 to 17percent); and largest when the San Francisco employment change relative to its suburbs wascompared to the Oakland employment change relative to its suburbs (22 to 29 percent).


15/34


TABLE 5

Wage and Employment Effects of City Minimum-wage Laws, Food Services






One year 0.03 0.07 0.06 0.05 0.01 0.03 0.01 0.02 0.01

(0.01)** (0.02)** (0.02)* (0.02)* (0.02) (0.03) (0.01) (0.02) (0.02)

Two years 0.06 0.06 0.06 0.02 0.02 -0.03 -0.01 -0.01 0.00

(0.01)** (0.02)* (0.02)** (0.03) (0.02) (0.04) (0.02) (0.02) (0.02)

Three years 0.07 0.07 0.07 -0.02 0.04 -0.04 -0.04 0.00 -0.03

(0.01)** (0.03)** (0.03)** (0.05) (0.02) (0.05) (0.02)* (0.03) (0.03)


One year 0.04 0.15 0.22 0.22 0.14 0.36 -0.12 -0.03 -0.06

(0.03) (0.11) (0.11)* (0.12)# (0.07)* (0.14)** (0.05)* (0.06) (0.07)

Two years 0.06 0.17 0.29 0.29 0.05 0.35 -0.18 0.07 0.03

(0.07) (0.13) (0.14)* (0.18) (0.12) (0.22) (0.07)* (0.08) (0.11)

Three years 0.01 0.17 0.26 0.33 -0.07 0.39 -0.22 0.10 0.12

(0.07) (0.13) (0.15)# (0.18)# (0.11) (0.21)# (0.08)** (0.17) (0.17)


One year 1.3 2.4 4.0 4.1 10.6 11.0 -16.5 -2.0 -8.8

Two years 1.0 3.1 4.5 15.7 2.0 -10.5 13.1 -10.6 -18.9

Three years 0.1 2.3 3.6 -19.1 -1.9 -9.2 5.7 22.7 -4.4


One year 0.2 0.6 1.0 0.4 0.2 0.7 -0.4 -0.1 -0.2

Two years 0.2 0.7 1.2 0.5 0.1 0.6 -0.7 0.3 0.1

Three years 0.0 0.6 1.1 0.5 -0.1 0.6 -0.9 0.5 0.6


The estimated wage and employment effects in panels (a) and (b) produce nine positive estimates ofthe labor demand elasticity (see panel (c)) and eight positive policy elasticities (see panel (d) andFigure 2). Among the policy elasticities, three were based on large, statistically significant changes inemployment relative to Oakland and its suburbs. The other policy elasticities were based on caseswhere the underlying employment increases were not statistically significant.


16/34


FIGURE 2

Employment-to-minimum-wage Elasticities, Food Services(a) San Francisco

s C

s s c c B B B

-1.2 -.9 -.6 -.3 0.0 .3 .6 .9 1.2

(b) Santa Fe

S

-1.2 -.9 -.6 -.3 0.0 .3 .6 .9 1.2 Notes: These histograms summarize all of the cases where the minimum-wage legislation was associated with anincrease in the average establishment wage in the industry (where the increase was statistically significant at at leastthe 10 percent level). If the minimum wage consistently lowered employment, then the estimated policy elasticitieswould all be to the left of the bar above zero. Elasticities based on own suburbs labeled S; reference city, C;reference city and its suburbs, B. Elasticities for year one, in red; year two, in blue; year three, in black. Elasticitiesin lower-case are not statistically significantly different from zero; estimates in upper case are significant at the 10percent level; in upper case and bold, at the 5 percent level or better. Estimates based on statistically insignificantincreases in average wages, or declines in wages, not shown. A maximum of nine possible elasticity estimates percity.Source: See Table 5.

Table 6 summarizes results for retail trade, another sector with a significant share of low-wageworkers. The San Francisco minimum wage, however, appears to have had little impact on wages inretail trade. Eight of the nine estimates of the wage impact hover near zero and only one relativeto wages in Oakland, in the third year after implementation shows a statistically significant increasein wages of about 4 percent (see panel (a)). Since the minimum wage was associated with an averagewage increase in only one of the nine cases, we have only one policy experiment to consider inpanel (b) the case of Oakland in the third year after the increase. The employment changes in theother eight cases correspond to situations where the minimum wage did not change relative wagesand therefore could not be expected to have an effect on relative employment. In the one casewhere San Francisco retail establishments saw their average wages increase in connection with theminimum wage, employment rose about 10 percent, but the increase was not statistically significant.The corresponding policy elasticity was 0.4 (see panel (d) and Figure 3).


17/34


TABLE 6

Wage and Employment Effects of City Minimum-wage Laws, Retail






One year 0.00 0.00 0.00 0.02 0.02 0.03 -0.02 -0.01 -0.01

(0.01) (0.01) (0.02) (0.02) (0.01)# (0.03) (0.02) (0.02) (0.02)

Two years 0.00 0.01 -0.01 0.06 0.02 0.09 -0.01 0.00 0.01

(0.02) (0.01) (0.02) (0.02)** (0.01)# (0.03)** (0.02) (0.02) (0.02)

Three years 0.01 0.04 0.01 0.05 0.03 0.07 0.01 0.00 0.02

(0.02) (0.02)* (0.02) (0.02)* (0.02)# (0.03)* (0.02) (0.02) (0.03)


One year 0.05 0.04 0.13 0.12 -0.04 0.22 -0.16 0.01 -0.02

(0.03)# (0.07) (0.07)# (0.09) (0.04) (0.11)* (0.05)** (0.07) (0.09)

Two years 0.00 0.09 0.21 0.21 -0.14 0.29 -0.25 -0.04 -0.05

(0.06) (0.14) (0.14) (0.13)# (0.07)# (0.15)# (0.06)** (0.08) (0.11)

Three years 0.00 0.10 0.18 0.30 -0.10 0.52 -0.42 -0.16 -0.21

(0.06) (0.15) (0.16) (0.16)# (0.10) (0.26)* (0.11)** (0.13) (0.16)


One year -18.3 13.5 -75.8 5.7 -2.1 7.7 10.4 -0.5 1.8

Two years -0.3 6.2 -41.0 3.5 -5.9 3.2 35.4 -13.3 -4.7

Three years -0.2 2.6 16.6 5.9 -3.8 7.8 -34.2 49.1 -12.5


One year 0.2 0.2 0.5 0.2 -0.1 0.4 -0.6 0.0 -0.1

Two years 0.0 0.3 0.8 0.3 -0.2 0.5 -0.9 -0.2 -0.2

Three years 0.0 0.4 0.7 0.4 -0.1 0.8 -1.5 -0.7 -0.8



18/34


FIGURE 3

Employment-to-minimum-wage Elasticities, Retail(a) San Francisco

c

-1.2 -.9 -.6 -.3 0.0 .3 .6 .9 1.2

(b) Santa Fe

c

C c S S B B

-1.2 -.9 -.6 -.3 0.0 .3 .6 .9 1.2

Notes: These histograms summarize all of the cases where the minimum-wage legislation was associated with anincrease in the average establishment wage in the industry (where the increase was statistically significant at at leastthe 10 percent level). If the minimum wage consistently lowered employment, then the estimated policy elasticitieswould all be to the left of the bar above zero. Elasticities based on own suburbs labeled S; reference city, C;reference city and its suburbs, B. Elasticities for year one, in red; year two, in blue; year three, in black. Elasticitiesin lower-case are not statistically significantly different from zero; estimates in upper case are significant at the 10percent level; in upper case and bold, at the 5 percent level or better. Estimates based on statistically insignificantincreases in average wages, or declines in wages, not shown. A maximum of nine possible elasticity estimates percity.Source: See Table 6.

The estimated labor-demand elasticities for San Francisco in panel (c) illustrate one pitfall ofattempting to interpret results in cases where there was not a meaningful policy experiment. Wagechanges that are small in economic magnitude and therefore also unlikely to be statisticallysignificantly different from zero can produce implausibly large positive or negative labor demandelasticities because the elasticity is the ratio of the employment change in panel (b) to the wagechange in panel (a). The estimated demand elasticities of 13.5 and -75.8, for example, are the productof plausible employment changes in panel (b) and very small (and statistically insignificant) wagechanges in panel (a).

Most earlier research on the minimum wage focused on industries (fast food) or particular groups ofworkers (teenagers) that are likely to be low wage. The QCEW data allow us to examine the wageand employment impacts of the minimum wage across all low-wage establishments, regardless of

industry or workforce composition. In Table 7 we display results from separate analyses of thebottom ( Table 7A ) and top ( Table 7B ) quintiles of the average establishment wage.13 We wouldexpect that the wage and employment effects of the minimum wage would be most concentratedamong establishments in the bottom quintile of the wage distribution; similarly, we would expect

13 We calculate quintiles separately for each of the three cities (together with their suburbs and the control cities andtheir suburbs).


19/34


that the minimum wage would have little or no impact on the average wage and employment inestablishments in the top quintile of the wage distribution.

TABLE 7A

Wage and Employment Effects of City Minimum-wage Laws, Bottom Wage Quintile






One year 0.03 0.10 0.10 0.11 0.05 0.11 -0.01 0.02 -0.02

(0.03) (0.04)* (0.05)# (0.03)** (0.02)** (0.03)** (0.02) (0.02) (0.03)

Two years 0.05 0.08 0.08 0.13 0.08 0.12 0.02 0.05 0.01

(0.03)# (0.03)** (0.04)* (0.03)** (0.03)** (0.04)** (0.02) (0.02)# (0.03)

Three years 0.11 0.12 0.10 0.14 0.10 0.15 0.00 0.05 -0.01

(0.04)** (0.04)** (0.04)* (0.05)** (0.03)** (0.05)** (0.03) (0.04) (0.03)


One year 0.01 -0.12 0.00 0.26 0.06 0.31 0.02 -0.20 -0.08

(0.10) (0.13) (0.15) (0.17) (0.13) (0.19)# (0.13) (0.20) (0.25)

Two years -0.01 -0.13 0.00 -0.35 -0.31 -0.45 -0.16 -0.26 -0.11

(0.10) (0.14) (0.16) (0.41) (0.26) (0.44) (0.12) (0.19) (0.23)

Three years -0.10 -0.10 -0.03 -1.46 -0.42 -1.63 -0.29 -0.32 -0.32

(0.11) (0.14) (0.16) (0.30)** (0.26) (0.34)** (0.10)** (0.17)# (0.21)


One year 0.3 -1.2 0.0 2.4 1.2 2.8 -2.1 -9.6 4.8

Two years -0.3 -1.7 0.0 -2.8 -3.9 -3.7 -7.5 -5.6 -13.0

Three years -0.9 -0.8 -0.3 -10.6 -4.3 -11.0 108.5 -6.5 42.1


One year 0.0 -0.4 0.0 0.5 0.1 0.6 0.1 -0.7 -0.3

Two years 0.0 -0.4 0.0 -0.4 -0.4 -0.5 -0.6 -1.0 -0.4

Three years -0.3 -0.3 -0.1 -0.9 -0.4 -1.0 -1.1 -1.2 -1.2

Notes: Authors' estimates using QCEW data. Ordinary least squares; standard errors in parentheses: **, 1percent; *, 5 percent; #, 10 percent. All regressions include controls for 19 industries and 4 establishment sizes.


20/34


TABLE 7B

Wage and Employment Effects of City Minimum-wage Laws, Top Wage Quintile






One year -0.03 -0.02 -0.02 -0.02 0.01 0.01 -0.01 -0.03 -0.04

(0.02)# (0.02) (0.02) (0.02) (0.01) (0.03) (0.03) (0.02) (0.03)

Two years -0.02 -0.02 -0.03 0.00 0.01 0.04 0.01 0.02 0.00

(0.02) (0.02) (0.03) (0.03) (0.02) (0.04) (0.01) (0.03) (0.02)

Three years -0.02 -0.01 -0.01 0.00 0.03 0.03 0.02 0.04 -0.01

(0.02) (0.03) (0.03) (0.05) (0.03) (0.06) (0.01) (0.03) (0.03)


One year 0.01 0.24 0.28 0.06 0.04 0.11 0.03 -0.08 0.04

(0.05) (0.18) (0.18) (0.08) (0.06) (0.17) (0.05) (0.09) (0.07)

Two years 0.06 0.22 0.23 0.13 0.05 0.14 0.02 0.08 -0.02

(0.07) (0.20) (0.18) (0.12) (0.07) (0.22) (0.07) (0.19) (0.09)

Three years 0.10 0.30 0.25 0.29 0.02 0.24 0.04 0.02 -0.05

(0.09) (0.22) (0.20) (0.19) (0.09) (0.28) (0.08) (0.20) (0.11)


One year -0.5 -9.6 -13.9 -2.9 3.8 14.8 -2.0 2.3 -1.1

Two years -2.9 -9.5 -8.8 -514.5 6.3 3.4 2.3 3.6 5.3

Three years -6.4 -22.3 -27.9 94.8 0.7 7.3 2.8 0.4 4.4


One year 0.1 1.0 1.2 0.1 0.1 0.2 0.1 -0.3 0.2

Two years 0.2 0.9 0.9 0.2 0.1 0.2 0.1 0.3 -0.1

Three years 0.4 1.2 1.0 0.4 0.0 0.3 0.2 0.1 -0.2

Notes: Authors' estimates using QCEW data. Ordinary least squares; standard errors in parentheses: **, 1 percent;*, 5 percent; #, 10 percent. All regressions include controls for 19 industries and 4 establishment sizes.

In San Francisco, the citywide minimum wage was associated with an economically important andstatistically significant increase in the average wage of low-wage establishments. The estimated wageincreases were between 3 and 12 percent, and were statistically significant at at least the ten percentlevel in eight of nine cases (see panel (a)). The corresponding employment changes, however, werenot significantly different from zero in any of the eight cases where wages rose (see panel (b)). Thecorresponding policy elasticities range from 0.0 to -0.4, but none are based on statistically significant

changes in employment (see panel (d) and Figure 4).


21/34


FIGURE 4

Employment-to-minimum-wage Elasticities, Lowest Wage Quintile Establishments(a) San Francisco

c c b b

c s s b

-1.2 -.9 -.6 -.3 0.0 .3 .6 .9 1.2

(b) Santa Fe

c

c

B S b s c s B

-1.2 -.9 -.6 -.3 0.0 .3 .6 .9 1.2

Notes: These histograms summarize all of the cases where the minimum-wage legislation was associated with anincrease in the average wage in establishments in the lowest quintile of the average establishment wage distribution(where the increase was statistically significant at at least the 10 percent level). If the minimum wage consistentlylowered employment, then the estimated policy elasticities would all be to the left of the bar above zero. Elasticitiesbased on own suburbs labeled S; reference city, C; reference city and its suburbs, B. Elasticities for year one, in red;year two, in blue; year three, in black. Elasticities in lower-case are not statistically significantly different from zero;estimates in upper case are significant at the 10 percent level; in upper case and bold, at the 5 percent level or better.Estimates based on statistically insignificant increases in average wages, or declines in wages, not shown. Amaximum of nine possible elasticity estimates per city.Source: See Table 7A.

As might be expected, the San Francisco minimum wage had no discernible effect on the average

wage of establishments in the top quintile of the areas wage distribution. In one case (in the firstyear after implementation, relative to the suburbs), the average wage appeared to fall, though theeffect was relatively small (3 percent) and only marginally significant.14

Some opponents of the San Francisco minimum-wage law expressed concerns that the legislationwould harm small employers. As a result, San Francisco exempted establishments with fewer than10 employees (as well as all nonprofit organizations) during the first two years that the newminimum-wage law was in effect.

In San Francisco (Table 8A), the minimum wage appears to have had little impact on average wageat any establishment size up to 100 employees. For the smallest firms (0-9 employees), which were

exempt from the new minimum for the first two years, the average wage change ranged from 0 to 1percent in the first two years, but none of these changes was statistically significant. In the third year when establishments with 0-9 employees became covered by the citywide minimum wage theaverage wage in these small firms rose a statistically significant 2 percent, relative to San Franciscosown suburbs, and about 3 percent relative to Oakland and Oakland and its suburbs, but these

14 In any event, there is no plausible story, in the context of either a standard competitive model or a dynamicmonopsony model by which raising the minimum wage would reduce wages.


22/34


slightly larger increases were not statistically significant. For the next size class (10-24) employees,which were covered from the initial implementation of the law, all the estimated wage effects aresmall (ranging from -1 to 1 percent) and none were statistically significant. For the third size class(25-100 employees), the estimated wage effects on these covered establishments were also not large(from -2 to 1 percent), mostly negative, and not statistically significant.15

TABLE 8AWage and Employment Effects of City Minimum-wage Laws, by Establishment Size, San Francisco

0-9 employees 10-24 employees 25-100 employees





One year 0.01 0.00 0.01 -0.01 0.00 0.00 -0.02 -0.02 -0.01

(0.01) (0.01) (0.01) (0.01) (0.01) (0.01) (0.01) (0.02) (0.02)

Two years 0.01 0.00 0.01 0.00 -0.01 0.00 -0.01 0.00 0.00

(0.01) (0.01) (0.01) (0.01) (0.01) (0.02) (0.01) (0.02) (0.02)

Three years 0.02 0.03 0.03 0.00 0.00 0.01 -0.01 0.01 0.00(0.01)* (0.02) 0.02 (0.01) (0.01) (0.02) (0.01) (0.02) (0.02)


One year 0.12 -0.07 0.10 -0.02 0.02 0.04 0.02 0.04 0.05

(0.05)* (0.09) (0.10) (0.02) (0.02) (0.03) (0.02) (0.03) (0.04)

Two years 0.13 -0.07 0.16 0.02 0.08 0.10 -0.01 0.05 0.05

(0.06)* (0.10) (0.11) (0.03) (0.03)* (0.04)* (0.03) (0.05) (0.05)

Three years 0.10 -0.04 0.14 0.05 0.14 0.13 -0.03 0.04 0.02

(0.08) (0.11) (0.13) (0.04) (0.04)** (0.05)** (0.04) (0.05) (0.06)


One year 17.7 -30.9 19.0 4.0 4.9 -8.8 -1.3 -1.9 -6.3Two years 13.3 -46.5 25.3 -3.9 -12.2 -38.7 0.9 -43.2 -24.0

Three years 5.5 -1.3 4.5 19.9 172.5 20.8 3.0 5.4 -21.6


One year 0.5 -0.3 0.4 0.0 0.0 0.1 0.1 0.1 0.2

Two years 0.5 -0.2 0.6 0.0 0.1 0.1 0.0 0.2 0.2

Three years 0.4 -0.1 0.5 0.1 0.2 0.2 -0.1 0.2 0.1

Notes: Authors' estimates using QCEW data. Ordinary least squares; standard errors in parentheses: **, 1 percent; *, 5percent; #, 10 percent. All regressions include controls for 19 industries.

Relative to San Francisco suburbs, employment increased in small, initially exempt, establishments inthe first two years after the increase, even though average wages in these establishments wereessentially unchanged. Employment remained above the suburban control group after the minimumwage began to apply to these same small establishments in the third year after the increase, thoughthe employment increase was not statistically significant.

15 We also conducted, but do not report results for larger establishment 100-999 and 1000 or more employees. Sincethere are relatively few firms in these size classes, particularly in the suburbs, we were not always able to produceestimates that fit our sample-size restrictions; results available upon request.


23/34


Relative wages increased in only one of the 21 potential policy experiments summarized in Table8B.16 Nevertheless, indirect effects may be at play here, with the minimum wage potentially drivingemployment into smaller exempt firms, discouraging smaller firms from growing larger, raising thegoing wage in the low-wage labor market regardless of the exemption, or through some other

channel. However, to the extent that there is an indirect effect, the small relative employmentchanges across all of the small establishment sizes suggest that these indirect effects were small.

TABLE 8B

Wage and Employment Effects of City Minimum-wage Laws, by Establishment Size, Santa Fe

0-9 employees 10-24 employees 25-100 employees





One year -0.01 0.00 0.01 0.02 0.01 0.01 0.06 0.03 0.08

(0.01) (0.01) (0.02) (0.02) (0.01) (0.02) (0.01)** (0.01)** (0.02)**

Two years 0.02 0.01 0.04 0.03 0.02 0.04 0.04 0.03 0.05(0.02) (0.01) (0.02) (0.03) (0.01)# (0.03) (0.02)# (0.01)* (0.03)

Three years 0.01 0.02 0.04 0.04 0.03 0.07 0.02 0.03 0.02

(0.02) (0.01) (0.03) (0.04) (0.01)* (0.04)# (0.04) (0.02)* (0.05)


One year 0.10 -0.07 0.23 0.11 -0.06 0.11 -0.05 0.02 0.07

(0.08) (0.05) (0.14)# (0.05)* (0.03)# (0.07) (0.07) (0.04) (0.13)

Two years -0.59 -0.35 -0.59 0.19 -0.09 0.19 0.05 0.05 0.21

(0.68) (0.16)* (0.71) (0.08)* (0.04)* (0.10)# (0.10) (0.05) (0.17)

Three years -3.21 -0.56 -3.26 0.13 -0.09 0.13 0.16 0.07 0.33

(0.66)** (0.20)** (0.70)** (0.10) (0.05)# (0.12) (0.13) (0.05) (0.19)#


One year -11.7 342.2 21.1 5.8 -4.3 7.5 -0.9 0.9 0.9

Two years -36.4 -35.4 -15.0 6.6 -3.8 5.3 1.2 2.1 4.6

Three years -278.7 -29.3 -90.8 3.1 -3.0 1.8 9.1 2.1 19.4


One year 0.4 -0.3 1.0 0.2 -0.1 0.2 -0.2 0.1 0.3

Two years -1.6 -1.0 -1.6 0.3 -0.1 0.3 0.2 0.2 1.0

Three years -3.4 -1.5 -3.4 0.2 -0.1 0.2 0.7 0.3 1.7

Notes: Authors' estimates using QCEW data. Ordinary least squares; standard errors in parentheses: **, 1percent; *, 5 percent; #, 10 percent. All regressions include controls for 19 industries.

Overall, the QCEW data show little support for the idea that the citywide minimum wageimplemented in San Francisco in 2004 cost the city jobs, either initially or up to three years after thelaw went into effect. In the overwhelming majority of cases where the minimum wage was

16 Nine cases each for establishments with 10-24 and 25-100 employees, and three in year three for establishmentswith 0-9 employees.


24/34


associated with increases in the average establishment wage in fast food, food services, retail, oramong all low-wage establishments (regardless of industry), employment did not change by astatistically significant amount. In those cases where employment did appear to respond in astatistically significant way to minimum-wage-related wage increases, employment actually increasedin four cases (all in food services) and declined in one (fast food). The San Francisco minimum wage

also appears to have had little or no effect on wages and employment at small establishments,including those that were initially exempt and those that were just above the initial cutoff.

SantaFe

In mid-2004, Santa Fe implemented a minimum-wage law that increased the prevailing minimumwage in the city about 65 percent in the first year after enactment, about 75 percent in the secondyear, and about 85 percent in the third year. Table 4 reports the wage and employment impact forfast-food restaurants. As panel (a) shows, in five of the nine cases considered, the citywide minimumwage was associated with statistically significant increases in fast-food wages of between 7 and 9percent (relative to fast-food wages in the three geographic controls); in four cases, wages alsoincreased by smaller amounts 1 to 3 percent but these changes were not statistically insignificant

levels; in a final case fast-food wages fell by a statistically insignificant amount (about 6 percent).None of the statistically significant wage increases, however, was associated with a statisticallysignificant change in employment (see panel (b)). Figure 1 displays the policy elasticities associatedwith each of the four cases where the average fast-food wage increased by a statistically significantmargin relative to the three geographic controls. Two were positive, two were negative, but nonewere based on statistically significant changes in employment. As was the case with San Francisco,the results for Santa Fe give little support to the view that the citys minimum wage hurtemployment in fast food.

Table 5 displays the results for the broader category of food services, which also includes sit-downrestaurants. The Santa Fe minimum wage appears to have had less impact on wages in food services

than in the narrower category of fast food. Only one of the nine cases studied in the first year afterimplementation, relative to the Santa Fe suburbs showed an increase in the wages of food-serviceestablishments. This estimated increase of about 5 percent was associated with a large, marginallysignificant increases in employment (see panel (b)). As Figure 2 shows, the implied policy elasticity inthis casewas 0.4. While the minimum-wage increase in Santa Fe did not have a large impact onaverage wages in food services, where it was associated with a wage increase, employment rose, notfell.

Table 6 extends the analysis to retail trade, where the city minimum wage appears to have had asubstantial impact on wages. In the first three years after the increase, the average wage rose by 2 to9 percent, depending on time period and geographic control, with seven of the nine changesstatistically significant or marginally significant (see panel (a)). In two of these cases of significant

wage increases, the corresponding employment changes were negative but not statisticallysignificantly different from zero; in four of these cases, employment increased after the wageincreases (with the increase significant at the ten percent level in three cases and the five percentlevel in the other case); and in one of these cases, employment fell (significant at the 10 percentlevel). Figure 3 summarizes the related policy elasticities. Four are positive, ranging from 0.3 to 0.8,and based on statistically or marginally significant increases in employment. Two are negative butnot statistically significantly different from zero. And one is negative, marginally significant, and at -0.2, in the range of historical estimates of disemployment effects.


25/34


In Table 7A, we broaden the analysis to include all low-wage establishments in Santa Fe, regardlessof industry. As in San Francisco, the Santa Fe minimum wage appeared to have an important impacton establishments in the lowest average-wage quintile before the increase. The estimated wageincreases were between 5 and 15 percent, and were statistically significant in all nine cases (see panel

(a)). Of these nine cases, the corresponding employment changes were positive and marginallysignificant in one case; negative and statistically significant in two cases; and statisticallyindistinguishable from zero in the remaining six cases (four negative and two positive). Figure 3shows the related policy elasticities.

As was the case with San Francisco, the Santa Fe legislation had no discernible effect on the averagewage of establishments in the top quintile of the areas wage distribution (see Table 7B).

Finally, we turn to the results by establishment size. In Santa Fe, the law exempted establishmentswith fewer than 25 employees.17 Not surprisingly, then, the increase did not appear to have muchimpact on the wages paid in establishments below the size cutoff. In the first year after the increase, wages in the smallest two establishment sizes we examined (0-9 and 10-24 employees) barely

changed relative to the three geographic control groups, with changes ranging from -1 to 2 percent,with none statistically significant (see panel (a) of Table 8B). In the second year after the increase,relative wages changes of 1 to 4 percent, though only one of these cases was marginally significant.In the third year, wages increased by statistically significant or marginally significant levels of 3 to 7percent in two of six cases, with statistically insignificant increases of 1 to 4 percent in the remainingcases. While the law exempted these smaller establishments, the small relative wage increases couldreflect changes in the Santa Fe labor market induced by the minimum-wage law. The city minimumwage, which raised the floor of the low-wage labor market well above the federal minimum wage,may have set a reference wage that affected pay even in exempted establishments. These kinds ofspillover dynamics, however, would not be consistent with the standard competitive model of thelow-wage labor market.18

Among covered small establishments those with 25 to 100 employees before the law went intoeffect wages increased by a statistically significant three to eight percent in six of nine cases (and bytwo to five percent in the three other cases where the changes were not statistically significant).Employment, however, did not change by a statistically significant amount in any of the six caseswhere wages rose. In five of the six cases, employment increased; in the remaining case, employmentfell. The corresponding policy elasticities ranged from -0.2 to 0.3.

17 The small-establishment exemption was initially intended to be permanent. At the end of 2007, the city council votedto remove the exemption for smaller establishments in 2008. During the full period analyzed here, establishments

with fewer than 25 workers were exempt from the Santa Fe minimum wage.18 If anything, in a competitive model, the minimum wage should lower average wages in exempt establishments

because employees who lost positions in larger, covered establishments would increase the labor supply available tosmaller, uncovered establishments, driving down the wage they have to pay to attract workers. One artefact of the

way the test is constructed may also explain the wage increase. Establishment size is determined by theestablishment's average number of employees in the year before the minimum wage went into effect. Establishmentsthat were initially below the 25-employee cutoff but that subsequently grew would be covered by the law after theincrease, potentially increasing their average wage.


26/34


The QCEW data by establishment size, therefore, suggest that the Santa Fe minimum wage raisedwages in covered small establishments without lowering employment. The data also hint that thelaw may have contributed to raising wages indirectly in smaller, exempt establishments.

Taken together, the results for Santa Fe, like those for San Francisco, provide little support for the

idea that the citywide minimum wage reduced employment opportunities in the low-wage labormarket there.

Washington,DC

Washington implemented its citywide minimum wage more than a decade before the laws put inplace in San Francisco and Santa Fe. On paper, the size of the increase a one dollar increase fromthe previously prevailing federal minimum wage of $4.25 per hour was about the same inpercentage terms as the increase in San Francisco. In practice, however, as weve already seen, theWashington minimum wage was set at a level that potentially affected fewer workers than in SanFrancisco or Santa Fe, and the subsequent compliance appears to have been so weak that the shareof workers earning below the new minimum wage actually increased after the law went into effect

(see Table 3).

The QCEW establishment data confirm the lack of a wage impact from the Washington minimumwage. In strong contrast to the experience of San Francisco and Santa Fe, the Washington minimumwage was associated with wage increases in only one of the 36 cases examined in fast food, foodservices, retail, and low-wage establishments (see Table 7A). The single case where wages appearedto rise for all low-wage establishments, in the second year after the increase, relative to the controlcity, Baltimore the increase was about five percent, but only marginally significant. In three cases,the new city minimum wage was associated with statistically or marginally significant declinesin fastfood or food service wages.

In the absence of binding minimum-wage increases, the QCEW data for Washington cannotprovide information on the employment impact of city minimum wages. The Washingtonexperience, however, does suggest the importance of setting a meaningful minimum wage in thecontext of the local pay structure and of providing adequate enforcement of the new wage level.

Discussion

Our analysis of city minimum wages covers multiple dimensions. We examine establishments indifferent industries, with different wage structures, and different initial sizes. We follow effects overperiods that vary from one to three yearsand we use three different geographic control groups for

each of the three cities studied. The qualitative results are remarkably consistent across all of thesedimensions. The minimum wages in San Francisco and Santa Fe, the two cities where the newminimum was set sufficiently high to affect wages in the low-wage labor market, had no systematiceffect on employment in affected establishments.

For San Francisco, Figure 5 pools all the cases for fast food, food services, retail, and all low-wageestablishments where the city minimum wage was associated with a statistically significant increase inthe average establishment wage. The figure groups the corresponding changes in employment into


27/34


five categories: a statistically significant decline in employment; a decline in employment that was notstatistically significant (at at least the 10 percent level); no change in employment; 19 an increase inemployment that was not statistically significant (at at least the ten percent level); and a statisticallysignificant increase in employment. In San Francisco, employment changes were almost evenlybalanced between negative and positive changes. Overall employment increases (14) outnumbered

decreases (11); statistically significant employment increases (3) also outnumbered statisticallysignificant employment declines (1). Figure 6 shows the corresponding results for Santa Fe. Again,employment changes were almost evenly divided between losses (11) and gains (10). Twice as manyobserved employment increases in Santa Fe were statistically significant at the 10 percent level orbetter (six) as employment declines (three).

FIGURE 5

Summary of All Employment Changes, San Francisco

1

10

2

11

3

0

5

10

15

Negative,

significant

Negative,

insignificant

Zero Positive,

insignificant

Positive,

significant

Numberofcases

Source: Authors' analysis of Tables 4, 5, 6, and 7.

19 In the figure, zero is defined as rounding to zero at one decimal place effectively covering policy elasticities inthe range -0.044 to 0.044.


28/34


FIGURE 6

Summary of All Employment Changes, Santa Fe

3

8

0

4

6

0

5

10

15

Negative,

significant

Negative,

insignificant

Zero Positive,

insignificant

Positive,

significant

Numberofcases

Source: Authors' analysis of Tables 4, 5, 6, and 7

The qualitative results in Figures 5 and 6 do not vary over time. Figure 7 combines the observedoutcomes in both San Francisco and Santa Fe and separates out the estimates by the first, second,and third year after implementation. For all three periods, the employment responses were aboutevenly distributed between negative and positive values. Year one shows five declines (onestatistically significant) and eight increases (three statistically significant); year two, eight declines(one significant) and eight increases (three significant); and year three, nine declines (two significant)and eight increases (three significant).

FIGURE 7

Summary of Employment Changes, San Francisco & Santa Fe

1

4

1

5

3

1

7

1

5

3

2

7

0

5

3

0

5

10

15

Negative,

significant

Negative,

insignificant

Zero Positive,

insignificant

Positive,

significant

Numberofcases

Year One

Year Two

Year Three

Source: Authors' analysis of Tables 4, 5, 6, and 7.


29/34


Nor do the qualitative results vary across the three geographic control groups. 20 The availability ofthree possible geographic controls allows us to observe the empirical variation across three plausiblecontrol groups, and to test the sensitivity of our quantitative and qualitative results to the choice of

control.

Figure 8 presents the full set of results for San Francisco (from Figure 5) according to the threegeographic controls: suburbs; nearby reference city; and the nearby reference city and its suburbs.Regardless of geographic control group, the qualitative employment effects are roughly evenlydistributed around zero. When the San Francisco suburbs were used as the reference group, there were three negative employment outcomes (one statistically significant), and five positiveemployment outcomes (all statistically insignificant). Relative to San Franciscos reference city(Oakland), employment fell six times (none statistically significant) and rose four times (nonesignificant). Finally, comparing San Franciscos performance relative to its suburbs with Oaklandsperformance relative to its suburbs, employment fell twice (neither significant), was unchangedtwice, and rose five times (three significant).

FIGURE 8

Summary of Employment Changes, San Francisco

1

2

0

5

00

6

0

4

00

2 2 2

3

0

5

10

15

Negative,

significant

Negative,

insignificant

Zero Positive,

insignificant

Positive,

significant

N

umberofcases

Suburbs

Reference City

Ref. City & Sub.

Source: Authors' analysis of Tables 4, 5, 6, and 7

Figure 9 displays the same data for Santa Fe. The qualitative results for Santa Fe appear to be

somewhat more sensitive to the choice of geographic controls than was the case for San Francisco.The employment outcomes using the Santa Fe suburbs and nearby reference city (Albuquerque)relative to its suburbs produce results that suggest, if anything, the Santa Fe minimum wage raisedemployment in the city. Relative to the Santa Fe suburbs, employment fell in two cases (onestatistically significant), but rose in six cases (three statistically significant). Using Albuquerque and

20 As Meyer (1995), Blundell, Costa Dias, Meghir, and Van Reenen (2004) and Draca, Machin, and Van Reenen (2011)and others have emphasized, the validity of the control group in difference-in-difference estimators depends on bothcommon trends and stable composition across the treatment and control groups.


30/34


its suburbs as a control, employment fell in two cases (one statistically significant) and increased inthree cases (all significant). Employment elasticities based on comparisons to Albuquerque alone,however, were more negative, with seven estimated declines (though only one was statisticallysignificant), with no estimates that were zero or positive.

FIGURE 9

Summary of Employment Changes, Santa Fe

1 10

3 3

1

6

01

01 1

0 0

3

0

5

10

15

Negative,

significant

Negative,

insignificant

Zero Positive,

insignificant

Positive,

significant

Numberofcases

Suburbs

Reference City

Ref. City & Sub.

Source: Authors analysis of Tables 4, 5, 6, and 7.

The policy elasticities in Tables 4 through 7 show substantial quantitative variation across geographiccontrols, even within the same city, industry, and time period. The qualitative results summarized in

Figures 8 and 9, however, suggest that the broad conclusions about the lack of employment impactsof the minimum wage do not depend on the choice of any of the three geographic controls.

Standard competitive models of the labor market cannot easily explain our results. Competitivemodels unambiguously predict that a binding minimum-wage will reduce employment. But, we findthis outcome in fewer than half the cases examined here, and in the large majority of those cases, theobserved employment declines were not statistically significant. One possible explanation is thatemployers adjusted to higher wage costs by reducing hours, not employment. Over the ranges seenhere, from the employees' point of view, hours adjustments yield an unambiguous welfareimprovement because average quarterly earnings are higher. If all the adjustment took place throughhours reductions, this implies that average hours fell while the average hourly wage rose by a larger

proportion, leaving employees working fewer hours for higher quarterly earnings.21

Other possible ways to reconcile our results with the competitive model include assuming that: labor costs arepassed on in higher prices (under the assumption that product demand is inelastic);22 minimum-wage

21 This result holds as long as the labor-demand elasticity is between 0 and -1.22 See Aaronson (2001), Fouger, Gautier, and le Bihan (2008), and Lemos (2008).


31/34


increases reduce non-wage compensation or training;23 and minimum wages reduce firm profits (butnot enough to drive firms out of affected industries).24

We could also reconcile our results with the standard competitive model if we assume that theminimum wage does lower employment but these disemployment effects are small. The random

distribution of negative and positive employment changes around a small, negative employmenteffect would be difficult to distinguish from the random distribution of negative and positiveemployment changes around an effect that is truly zero. From a policy point of view, the differencebetween zero and small negative employment effects is not likely to be important, especially in acontext where the wage benefits are large.

In general, our results are more consistent with non-competitive recruitment-retention models ofthe labor market.25 These models differ from the competitive model in that they assume that firmsface costs for recruiting and retaining workers. In this context, an increase in the minimum wage canactually make it easier for firms to recruit and retain workers, offsetting some or all of the increasedlabor costs. As a result, the recruitment-retention model suggests that over some range, minimum-wage increases might result in no change or even an increase in employment.

Conclusion

The results for fast food, food services, retail, and low-wage establishments in San Francisco andSanta Fe support the view that a citywide minimum wages can raise the earnings of low-wageworkers, wi

The Wage and Employment Impact of Minimum-Wage Laws in Three Cities

Documents

Transcript of The Wage and Employment Impact of Minimum-Wage Laws in Three Cities